Lehmbach and Krummenohl Wastewater Treatment Plants

The Lehmbach and Krummenohl wastewater treatment plants are among the just under 40 wastewater treatment plants belonging to Aggerverband.

Both plants functioned as pilot projects, i.e. the predictive control system developed by the GECO►C team was put into operation at them and studied in detail. This system of control is based on artificial neural networks, so-called ANNs.

Prior to the control system being put into operation, the plants exhibited typical problems such as ammonium peaks within the outflow and/or increased energy costs. Following installation of the controller, it was able to be shown in a long-term study that there was a clear reduction in the outflow peaks as well as a reduction in energy consumption. In addition, in the case of the Krummenohl wastewater treatment plant, the impending expansion of the plant was able to be stopped.

"Schematically installation of control strategy with ANN"

The principle of the control system consists in the historical data of the plant being analysed with the aid of artificial neural networks. This subsequently results in an acquired ANN that is able to produce forecasts of outflow values in an hour, for example. The forecasts were used to produce set values for the subordinated control system (e.g. controlling the oxygen during the aeration phase) with the aid of a fuzzy controller. The fuzzy controller, as a multiparameter controller, is able, not only to react to the current measured values but also to take into account those of the future. Thus, the controller can react early in the event of any shock loads and better exploit existing capacity reserves. In addition, it was possible to reach a stage, through the integrated, continuous relearning process of the ANNs, where the control of the plant, even in the event of a change in operating behaviour, adjusts itself continuously.

The controller at the Krummenohl WWTP, based on artificial neural networks, was converted in order to match the set value, dependent on the load, for controlling the oxygen during the aeration phase.